NR

Image Compression Using

the Haar Wavelet Transform
Peggy Morton and Arne Petersen

2004. 10. 18
Yoon HeeJoo

1
Contents
Introduction
Averaging and Differencing
Image Representation
Using Linear Algebra
Image Compressions

2
1. Introduction
Done by compressing the image
Use a process called averaging and
differencing to develop a new matrix
representing the same image in a
more concise manner.
Eliminate some of unnecessary
information, and arrive at an
approximation of our original image

3
2. Averaging and Differencing
(1/3)
8X8 matrix, process involve three steps
(2=8)
Step 1
[3 5 4 8 13 7 5 3]
Averaging

Differencing

[4 6 10 4 -1 -2 3 1]
detail coefficient

ex. (3 + 5)/2 = 4
3 - 4 = -1
4
2. Averaging and Differencing
(2/3)
Step 2
[4 6 10 4 -1 -2 3 1]

Averaging

Differencing

[5 7 -1 3 -1 -2 3 1]
detail coefficient

ex. (4 + 6)/2 = 5
4 - 5 = -1

5
2. Averaging and Differencing
(3/3)
Step 3
[5 7 -1 3 -1 -2 3 1]
Averaging
Differencing

[6 -1 -1 3 -1 -2 3 1]
detail coefficient
row average

ex. (5 + 7)/2 = 6
5 - 6 = -1
6
3. Image Representation (1/6)
Use the averaging and differencing
process 64 2 3 61 60 6 7 57
9 55 54 12 13 51 50 16

17 47 49 20 21 43 42 24

40 26 27 37 36 30 31 33
A
32 34 35 29 28 38 39 25

41 23 22 44 45 19 18 48
49 15 14 52 53 11 10 56

8 58 59 5 4 62 63 1

[Figure 1]

7
3. Image Representation (2/6)
first row
[64 2 3 61 60
6 7 57]

Step1 [33 32 33 32 31 -29

27 -25]
Step2 [32.5 32.5 0.5 0.5 31 -29
27 -25]
Step3 [32.5 0 0.5 0.5 31 -29
27 -25]
8
3. Image Representation (3/6)
rows the results
32.5 0 0.5 0.5 31 29 27 25
32.5 0 0.5 0.5 23 21 19 17

32.5 0 0.5 0.5 15 13 11 9

32.5 0 0.5 0.5 7 5 3 1
32.5 0 0.5 0.5 1 3 5 7

32.5 0 0.5 0.5 9 11 13 15
32.5 0 0.5 0.5 17 19 21 23

32.5 0 0.5 0.5 25 27 29 31
detail coefficients
row average
9
3. Image Representation (4/6)
columns the results
choose some number () and set
equal 0 5to
32.5 32. 00 zero
0 0 0 00 00 00 00
0 00 00 00
00 0 00 00 00
0 00 00 00 0 40 04 40 04

0 00 00 00 0 40 04 40 04
0 00 0.05 00.50 27
27 2525 23
23 2121

0 00 .05 0 .50 1111 99 77 50
0 00 0.05 00.50 05 77 99 11
11

0 00 .05 0 .50 21
21 2323 25
25 2727 =5
10
3. Image Representation (5/6)
apply the inverse of the averaging
the differencing operations
59.5 5.5 7.5 57.5 55.5 9.5 11.5 53.5
5.5 59.5 57.5 7.5 9.5 55.5 53.5 11.5

21.5 43.5 41.5 23.5 25.5 39.5 32.5 32.5

43.5 21.5 23.5 41.5 39.5 25.5 32.5 32.5
32.5 32.5 39.5 25.5 23.5 41.5 21.5 43.5

32.5 32.5 25.5 39.5 41.5 23.5 43.5 21.5
53.5 11.5 9.5 55.5 57.5 7.5 5.5 59.5

11.5 53.5 55.5 9.5 7.5 57.5 59.5 5.5

11
3. Image Representation (6/6)

Original Image Decompressed Image

[Figure 2]

12
4. Using Linear Algebra (1/3)
Step1 1
2 0 0 0
1
0 0 0

2
1 1
0 0 0 0 0 0
2 2
1 1
0 2
0 0 0
2
0 0
1 1
0 0 0 0 0 0
A1 2
1
2
1

0 0 0 0 0 0
2 2
1 1
0 0 0 0 0 0
2 2
0 1 1
0 0 0 0 0
2 2
1 1
0 0 0 0 0 0
2 2

Average Difference

13
4. Using Linear Algebra (2/3)
Step2 Step3
1 1
2 0 0 0 0 0 0 1 1
2 2 0 0 0 0 0 0
1 1 2
0 0 0 0 0 0 1 1
2 2 0 0 0 0 0 0
1 1 2 2
0 2
0
2
0 0 0 0 0 0 1 0 0 0 0 0
A2 1 1 A3 0

0 0 1 0 0 0 0

0 0 0 0 0 0
2 2 0 0 0 0 1 0 0 0
0 0 0 0 1 0 0 0
0 0 0 0 0 0 1 0 0
0 0 0 0 1 0 0 0
0 0 0 0 0 1 0
0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 1
0 0 0 0 0 0 0 1

14
4. Using Linear Algebra (3/3)
1 1 1 1
W = A1A2A3A 8 8 4
0
2
0 0 0
1 1 1 1
0 0 0 0
8 8 4 2
1 1 1 1
8 0 0 0 0
8 4 2
1 1 1 1
0 0 0 0
8 8 4 2
1 1 1 1
8 0 0 0 0
8 4 2
T 1
T W AW
8

1
8
0
1
4
0 0
1
2
0

1 1 1 1
T 1 1 0 0 0 0
(W ) AW A 8
1

8
1
4
1
2
1
0 0 0 0
8 8 4 2
A : Original Matrix
W : Transforming
Matrix
T : Compressed
Matrix
15
5. Image Compressions

Figure 3a. Figure 3b. Figure 3c.

65,536 entries 65,536 entries
Original
631 non-zero 1,411 non-zero
Image entries entries
Compression Compression
Ratio 103:1 Ratio 46:1

16
Thank you for Attending My
Presentation
Program

Questions

Additional Explanations

17